Reactive Power Allocation Method in a Wind Farm for ... · Keywords – Wind farm operation,...

International Journal of Engineering Research and Technology.

ISSN 0974-3154 Volume 11, Number 12 (2018), pp. 2167-2182

© International Research Publication House

http://www.irphouse.com

Reactive Power Allocation Method in a Wind Farm

for Improved Voltage Profile and Loss Reduction

Hong-Chao Gao, Sang-Yun Yun, Joon-Ho Choi, Seon-Ju Ahn*

Department of Electrical Engineering, Chonnam National University,

Republic of Korea.

*Corresponding Author

ABSTRACT: Recently, the reactive power control for wind farm is required with the

rapidly increasing integration of wind energy. In general, the two main

objectives of the reactive power control are the power loss reduction and

offering rapid reactive power support to power grid during fault. This paper

focuses on the steady-steady reactive power control of wind turbine

generators to achieve the economical operation of a wind farm. It aims to

improve the voltage profile within a wind farm to reduce the action of

voltage regulation equipment, thereby reduces operation costs. Meanwhile,

the conventional control objective, like power loss reduction, is also taken

into consideration. To achieve the above goals, four optimal reactive power

allocation methods are proposed and formulated as a quadratic

programming problem using the linearized relationship between the voltage

and reactive power. To verify the performance, a comprehensive wind farm

model with 40 WTGs is developed for simulation. The effectiveness of the

proposed reactive power allocation method is verified compared with the

conventional even allocation method. Moreover, the simulations with 100

different scenarios considering the reactive power reference and wind speed

are conducted to demonstrate the feasibility and reliability.

Keywords – Wind farm operation, Reactive power allocation, OLTC,

Loess reduction, Voltage profile improvement

I. INTRODUCTION

Wind energy is one of the major renewable energy sources and more and more wind

farms have been constructed. According to the world wind farm database,

approximately 19,000 wind farms are constructed in the world with total capacity of

675.6 GW [1]. As the wind power capacity has been increasing, power system also

faces with a lot of challenges, such as power quality, voltage variation, voltage dip,

2168 Hong-Chao Gao, Sang-Yun Yun, Joon-Ho Choi, Seon-Ju Ahn

harmonics and flickers and so on [2]. Benefitting from the modern wind turbines and

power electronic technologies, reactive power control is the major solution to fulfill the

requirements of dynamic voltage stability described in grid codes. Reference [3]

described reactive power management of wind farm in most technical and economical

way considering the wind turbine technology. The significance of reactive power

control in the wind farm has been reported in many literatures. For example, wind

turbine generators (WTGs) are generally required to have the ability to subject to high

voltage ride through and low voltage ride through during grid fault [4-5]. From the

viewpoint of system operation, reactive power control of wind farm is also used to

reduce the power losses and improve the voltage profile [6].

From the perspective of means to implement reactive power control of wind farm, static

synchronous compensators (STATCOM), static VAR compensator (SVC) and other

devices are showing excellent control performance. For example, STATCOM provides

better damping characteristics, which is the best suited for dynamic stability [7].

Moreover, SVC is able to regulate the voltage and stabilize the system and it can bring

the system closer to unity power factor. However, it is not suitable for the case of high

wind power generation [8]. In addition, these devices greatly increase the wind farm

costs. Therefore, it is a highly feasible and economical solution to utilize fully the

inherent reactive power control capability of WTGs. However, distributed and local

control of wind farm can make configuration complex, and has a big difficulty in

following the grid code requirements. Thus, many researcher proposed centralized

control strategy [9-10]. In the centralized control strategy, the principle to allocate the

reactive power requirement at the point of common coupling (PCC) and designate the

reference signal to each single WTG in the wind farm is very important. In fact, optimal

reactive power allocation is not only mentioned for wind farm control, but also for the

conventional distribution system. For example, a novel allocation method of reactive

power, which takes the production cost and transmission cost into consideration at the

same time, is proposed in [11]. Similarly, several reactive power allocation methods for

wind farm are also proposed in [12-15]. In [12], even allocation is proposed where each

wind turbine generator will be controlled with the average reactive power reference

value. It is very simple to implement. However, every WTG in the wind farm may not

has the same operation state, thus even allocation cannot fully take into account the

capability of individual WTGs and hardly achieve economical operation of wind farm.

The proportional allocation, where each WTG will be designated the reactive power

reference value by proportion principle referring to the capacity limit of each WTG, has

been proposed in [13]. By this method, the voltage profile of wind farm has not been

taken into consideration, thus the burden of voltage regulating devices is increasing.

There are also several allocation methods with the objective of minimizing the power

loss, but the lifetime cost of devices is not considered [14-15].

The economical operation of wind farms is undoubtedly a key factor. Generally, the

OLTC is an appropriate means to achieve the goal of regulating the voltage of wind

farm. As it is a mechanical device, its lifetime directly affects economic efficiency. The

rational and improved voltage profile in wind farm can reduce the burden on these

devices. Therefore, improvement of voltage profile is imperative. Benefitting from the

Reactive Power Allocation Method in a Wind Farm for Improved Voltage Profile and Loss Reduction 2169

close relationship between the terminal voltage and reactive power flow of WTG, an

appropriate reactive power allocation and control means is enough to achieve the goal

of improvement of voltage profile. Herein, the primary focus is reducing the number of

tap changer operation through improving voltage profile within wind farm by using a

novel reactive power allocation method. Meanwhile, the power loss is also considered.

This is formulated as a quadratic programming problem, where the goals of OLTC

operation times reduction and power loss reduction is not formulated into objective

function directly. Instead, the relationship between voltage and reactive power is

linearized and fully utilized.

The remaining part of this paper is organized as follows: a developed offshore wind

farm model with 40 WTGs is descripted in detail in section 2 and it may provide some

parameter reference for other researchers, and section 3 discusses and analyzes the

proposed reactive power allocation methods and their formulations. In section 4, the

simulation results illustrate the feasibility and effectiveness of proposed reactive power

allocation method. Moreover, different wind directions’ wake effects are also

considered and analyzed. Finally, section 5 will draw the conclusion for the research.

II. Wind Farm Model

II.I Wind Farm Configuration

Wind farm can be modelled differently according to the objective of the research. When

the effect of wind farms on the transmission system is the main concern, wind farm is

usually represented by one aggregated generator driven by a single equivalent wind

turbine [16-17]. On the contrary, a complete wind farm model with an exact number of

WTGs is preferred for wind farm controller design studies [12]. In this paper, a detailed

wind farm model composed of 40 GE 3.6MW wind turbines is used. The configuration

and the parameters, including bus numbers, are shown in Fig 1. The wind farm consists

of five radials, where eight WTGs are installed in each radial. The 33kV cables are used

to connect each radial to the offshore platform bus. The offshore transformer installed

at offshore platform is used to step up the voltage from the 33kV of wind farm to 150kV

of the submarine cable for high voltage transmission. The data for the offshore

transformer are summarized in Table 1.

II.II Offshore Transmission System

Many large scale offshore wind farms have been built in the last decade, such as Horns

Rev of Denmark, North Hoyle of the UK and so on [18]. For large wind farms, offshore

substations are required for stepping up the voltage level and for converting the power

to HVDC. The choice of appropriate design and technology for the transmission system

can be a crucial factor of the offshore wind farm projects [19]. In this paper, a 20 km

HVAC transmission is assumed which is based on the report from the real wind farm

projects [18]. The HVAC cable adopts 150 kV XLPE and its technical data are

summarized in Table 2.


PL

AT

FO

RM

TR

AN

SFO

RM

ER

3

3/1

50 [k

V]

Row

Column 1 2 3 4 5 6 7 8

1

2

3

4

5

2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17

363534

3332313029282726

2524232221201918

4140393837

33 kV XLPE 400 mm2

240 mm2 120 mm2 95 mm2

150 kV XLPE

Sh. R8MVar

Sh. R8MVar

0

90

180

270

Wind direction

Fig. 1. Wind farm configuration

The generation of large amounts of reactive power is a major limiting factor in the use

of HVAC cables in long distance transmission systems [19-20]. Therefore,

consideration of the effect of reactive power generation induced by HVAC is necessary.

An estimation of the reactive power generation by the 150 kV HVAC can be calculated

based on equation (1) [20]

𝑄 = 𝜔 × 𝐶 × 𝑙 × 𝑉2 (1)

where 𝜔 is the angular frequency, 𝐶 is the capacitance, 𝑙 is the length and 𝑉 is the

voltage. In fact, compensation only at the onshore end is possible, but adding

compensation both sides can greatly improve the current profile along the HVAC link,

consequently transmission loss can be reduced. Therefore, two 8 MVar shunt reactors

are installed at both ends to compensate for the 60% of the estimated reactive power

generated in the cable.

II.III Wind Farm Cable Selection

Larger conductor cross-section gives less loss and higher power rating, but it is more

expensive. When sizing cables, it is often preferred to size the largest cable first and

choose the size for the intermediate and small cables in sequence. In this wind farm

model, the largest cables are connected between the offshore platform and the first wind

turbine of each radial by using 400 mm2 cables. The length of the cables from the

offshore platform to each radial is shown in Fig 1, and technical data are shown in Table

3. Based on the power rating, the cable in each radial is divided into three different sizes

[21]. The first two cables from column 1 to 3 adopt a cross section of 240 mm2, and

120 mm2 cables are adopted for the cables from column 3 to 5. The last three cables

were selected to support three wind turbines with the 95 mm2 cross section.

The distance between each wind turbine is assumed to be six rotor diameters, where the

rotor diameter is 104 m. The distance from seabed to the wind turbine is 100 m and the

cable length at seabed should include 3% slack of the distance at seabed [21]. Thus, the

length of the cables between any two wind turbines is 0.8427 km. The impedance data

for the cables in each radial are given in Table 4.


III. Reactive Power Allocation Methods

Fig 2 illustrates the reactive power allocation scheme based on the centralized control

strategy for a wind farm. First, the reactive power requirement at the PCC (𝑄𝑊𝐹∗ ) is

determined by the transmission system operator (TSO). After that, the wind farm

controller allocates the reactive power requirement and dispatches it to each WTG.

Herein, the simplest and conventional allocation method, even allocation method, will

be used as benchmark [12], which are further elaborated based on equation (2)

𝑄𝑟𝑒𝑓,𝑊𝑇𝑖=

1

𝑁𝑊𝑇𝑄𝑊𝐹

∗ (2)

where, 𝑄𝑟𝑒𝑓,𝑊𝑇𝑖 is the reactive power reference value of each WTG received from wind

farm controller and 𝑁𝑊𝑇 is the total number of WTGs in a wind farm. As mentioned in

section 1, the voltage profiles of WTG buses are mainly dependent on the active and

reactive power flow of each WTG. Therefore, the steady state voltage profiles of WTGs

can be improved by proper reactive power allocation. With improved voltage profiles,

the number of switching operation of tap changing transformer and/or capacitors and

reactors can be reduced and the power loss in the wind farm can also be decreased.

Considering the above mentioned purposes, four reactive power allocation methods are

proposed and analyzed.

Table 1. Offshore transformer data

Bus

From

Bus

To

Rated voltage

[kV]

Power rating

[MVA]

Rated impedance

[%]

1 42 33/150 160 13.8

Table 2. 150 kV HVAC XLPE cable data

Bus number Length [km] R [] L [ohm] C [uF]

42-43 20 0.78 2.40 3.80

Table 3. Data for the cables between the offshore platform and the radials

Bus number Length [km] R [ohm] L [ohm] C [uF]

1-2/1-34 12.56 0.7536 1.3810 3.5168

1-10/1-26 7.41 0.4460 0.8148 2.0748

1-18 2.26 0.1356 0.2485 0.6328

Table 4. Data for the cables between wind turbines

Cross-Section

[ mm2]

Power rating

[MVA]

R

[ohm]

L

[ohm]

C

[uF]

95 15.8 0.2023 0.11643 0.13483

120 18.6 0.1685 0.10857 0.15169

240 29.3 0.0843 0.09796 0.19383


Grid busOnshore

bus

PCC

WF ControllerAllocation method

IWTrefQ ,IWTrefQ ,

Pla

tform

High voltage tranmission

TSOControl center

*WFQ

Fig. 2. Schematic of centralized wind farm reactive power allocation

III.I Reactive Power Allocation Method 1

The objective of the method 1 is to make the terminal voltage of each WTG as close to

the wind farm average voltage value (𝑉𝑎𝑣𝑔) as possible. Because of the impedance of

the transmission line in wind farm, the terminal voltage of WTGs along any radial line

will show a monotonically increasing profile. The objective of method 1 is to make the

voltage profile in each radial as flat as possible. The OLTC installed at offshore

platform is used to regulate the voltages of wind farm. If the voltage of any point

violates, the OLTC will change its tap position to pull it back to allowable range.

Therefore, an improved voltage profile can reduce the number of OLTC operation,

which can lengthen its lifetime, thus economic effectiveness can be achieved. The

process of reactive power allocation to each WTG can be formulated as follows. The

objective function is presented as equation (3)

min ∑ (𝑉𝑖 − 𝑉𝑎𝑣𝑔)2𝑁𝑊𝑇

𝑖=1 (3)

where, 𝑉𝑖 is the terminal voltage of the ith WTG.

The constraints are given in equations (4) – (6).

∑ 𝑄𝑗 = 𝑄𝑊𝐹 ∗𝑁𝑊𝑇

𝑖=1 (4)

𝑄𝑗𝑚𝑖𝑛 ≤ 𝑄𝑗 ≤ 𝑄𝑗

𝑚𝑎𝑥 (5)

𝑉𝑖,𝑚𝑖𝑛 ≤ 𝑉𝑖 ≤ 𝑉𝑖,𝑚𝑎𝑥 (6)

The sum of reactive power reference signal (𝑄𝑗) allocated to each WTG should be equal

to the reactive power requirement (𝑄𝑊𝐹∗ ) at PCC to meet the grid requirements. The

voltages of WTGs should be maintained within the upper (𝑉𝑖,𝑚𝑎𝑥) and lower (𝑉𝑖,𝑚𝑖𝑛)

limits, and the reactive power reference signal for each WTG should be within the

capacity limits (𝑄𝑗𝑚𝑎𝑥 and 𝑄𝑗

𝑚𝑖𝑛).

In order to solve the above optimization problem, the linearized relationship between

the reactive power and the voltage, given in equation (7), is used

∆𝑉𝑖 = 𝑉𝑖 − 𝑉𝑖,0 = ∑ Z(𝑖, 𝑗)∆𝑄𝑗𝑁𝑊𝑇𝑗=1 (7)


where, Z(𝑖, 𝑗) is the element of the sensitivity matrix, and represents the sensitivity

between the voltage at bus i and the reactive power value of the WTG j. ∆𝑉𝑖 represents

the voltage change from the initial voltage (𝑉𝑖,0) after the reactive power control. The

constraints (4) – (6) can be transformed into equations (8) – (10)

∑ ∆𝑄𝑗 = ∆𝑄𝑊𝐹∗𝑁𝑊𝑇

𝑗=1 (8)

∆Qjmin ≤ ∆Qj ≤ ∆Qj

max (9)

𝑉𝑖,𝑚𝑖𝑛 − 𝑉𝑖,0 ≤ ∑ 𝐙(𝑖, 𝑗)∆𝑄𝑗

𝑁𝑊𝑇

𝑗=1≤ 𝑉𝑖,𝑚𝑎𝑥 − 𝑉𝑖,0 (10)

where, ∆𝑄𝑊𝐹∗ is the variation of the reactive power requirement at PCC for the wind

farm. ∆𝑄𝑗 is the reactive power variation of WTG j.

The problem is formulated as a quadratic programming problem whose general form is

given in (11) – (14).

min 1

2𝑥𝑇𝐻𝑥 + 𝑓𝑇𝑥 (11)

𝐴 ∙ 𝑥 ≤ 𝑏 (12)

𝐴𝑒𝑞 ∙ 𝑥 = 𝑏𝑒𝑞 (13)

𝑙𝑏 ≤ 𝑥 ≤ 𝑢𝑏 (14)

Reactive power reference values designated to each WTG are defined as x in equation

(15) and the matrix H and vector f of the objective function are given by equations (16)

and (17).

𝑥 = [∆𝑄1, ∆𝑄2, ⋯ , ∆𝑄𝑁𝑊𝑇]

𝑇 (15)

𝐻(𝑚, 𝑛) = ∑ 𝐾(𝑘, 𝑚) ∗ 𝐾(𝑘, 𝑛)𝑁𝐵𝑈𝑆𝐾=1 (16)

𝑓(𝑚) = ∑ 𝐾(𝑘, 𝑚) ∗ (𝑉𝑘,0 − 𝑉𝑎𝑣𝑔,0)𝑁𝐵𝑈𝑆𝑘=1 (17)

Where, 𝑉𝑎𝑣𝑔,0 is the initial average voltage. K is the modified sensitivity matrix, and it

is used to represent the sensitivity between the voltage at bus 𝑖 and the reactive power

value of the WTG 𝑗 whose elements are calculated form the matrix Z as follows.

𝐾(𝑖, 𝑗) = 𝑍(𝑖, 𝑗) −1

𝑁𝑊𝑇∑ 𝑍(𝑖, 𝑗)𝑁𝑊𝑇

𝑖=1 (18)

The inequality constrains are given as follows.

𝐴(𝑖, 𝑗) = {𝑍(𝑖, 𝑗) 1 ≤ 𝑖 ≤ 𝑁𝑊𝑇

−𝑍(𝑖 − 𝑁𝑊𝑇 , 𝑗) 𝑁𝑊𝑇 + 1 ≤ 𝑖 ≤ 2𝑁𝑊𝑇 (19)

𝑏(𝑖) = {𝑉𝑚𝑎𝑥 − 𝑉𝑖,0 1 ≤ 𝑖 ≤ 𝑁𝑊𝑇

𝑉(𝑖−𝑁𝑊𝑇),0 − 𝑉𝑚𝑖𝑛 𝑁𝑊𝑇 + 1 ≤ 𝑖 ≤ 2𝑁𝑊𝑇 (20)

Finally, the equality constrains are represented as follows.

𝐴𝑒𝑞 = [1,1, ⋯ ,1] (21)


𝑏𝑒𝑞 = ∆𝑄𝑊𝐹 ∗ (22)

III.II Reactive Power Allocation Method 2

The objective function of the proposed allocation method 2 is the same as that of the

method 1, which makes the terminal voltages of WTGs as close to the average voltage

of wind farm as possible. However, in this method, the reactive power limit of each

WTG is set according to the sign of the wind farm reactive power reference. If the wind

farm is required to supply the reactive power to the grid, i.e. 𝑄𝑊𝐹∗ > 0, each WTG is

required not to absorb the reactive power. In the opposite case, all WTGs are required

not to supply the reactive power. This is because that the improvement of the voltage

profile by using allocating reactive power may induce unnecessary reactive power

current along the cable. Without this restriction, unnecessary reactive current flowing

between WTGs can increase power loss. The formulations are the same as those

mentioned in the method 1, except the constraint (5) is changed to (23).

{0 ≤ 𝑄𝑗 ≤ 𝑄𝑗

𝑚𝑎𝑥 𝑄𝑊𝐹∗ ≥ 0

𝑄𝑗𝑚𝑖𝑛 ≤ 𝑄𝑗 ≤ 0 𝑄𝑊𝐹

∗ < 0 (23)

III.III Reactive Power Allocation Method 3

Allocation method 3 is proposed to try to make the terminal voltages of WTGs closer

to the upper limit (Vmax). As well known, the power loss in a wind farm has a close

relationship with the impedance and current of the cable. Due to the inverse relationship

between voltage and current, higher terminal voltage decreases the cable current,

thereby reducing the power loss [22]. The objective function can be formulated as

equation (24), while the constraints are the same as equations (4), (6), and (23)

min ∑ (𝑉𝑖 − 𝑉𝑚𝑎𝑥)2𝑁𝑊𝑇𝑖=1 (24)

where, 𝑉𝑚𝑎𝑥 is the maximum limit of the voltage in a wind farm. The equality and

inequality constraints are identical to the equations (19) - (22). The matrix H and vector

f of the objective function for this allocation method are given as follows.

𝐻(𝑚, 𝑛) = ∑ 𝑍(𝑘, 𝑚) ∗ 𝑍(𝑘, 𝑛)𝑁𝐵𝑈𝑆𝐾=1 (25)

𝑓(𝑚) = ∑ 𝑍(𝑘, 𝑚) ∗ (𝑉𝑘,0 − 𝑉𝑚𝑎𝑥)𝑁𝐵𝑈𝑆𝑘=1 (26)

III.IV Reactive Power Allocation Method 4

Allocation method 4 aims to coordinate the two concerned goals: to reduce the number

of OLTC tap operation and to reduce the power loss. To further investigate the voltage

profile and power loss in a wind farm, Fig. 3 illustrates five representative voltage

profiles in a radial according to the reactive power control of WTGs. Line (a) represents

a voltage profile when all WTGs operate in a unity power factor, where the voltage rise

along the line is due to the active power of the WTGs. When the foremost WTG

supplies or absorbs reactive power, the voltage rise/drop at each WTG terminal will be


almost the same. Therefore, the lines (b) and (d) are parallel to line (a). On the contrary,

when the hindmost WTG supplies or absorbs reactive power, the voltage rise/drop at

each WTG terminal will be proportional to the length from the platform transformer, as

presented in lines (c) and (e).

If the WTGs installed at the end of the radial supply or absorb the reactive power, the

reactive current in the line will increase the power loss. Therefore, by allocating the

required reactive power to the upstream WTGs, power loss due to the reactive current

can be reduced. When 𝑄𝑊𝐹∗ > 0, and the upstream WTGs supply the required reactive

power, the voltage profile will be similar to line (b). On the contrary, the desired voltage

profile will be like line (b) if 𝑄𝑊𝐹∗ < 0. From the above analysis, a coordinated rule for

combining the methods 2 and 3 according to the sing of 𝑄𝑊𝐹∗ is proposed as follows:

{𝑃𝑟𝑜𝑝𝑜𝑠𝑒𝑑 𝑚𝑒𝑡ℎ𝑜𝑑 2 𝑄𝑊𝐹

∗ > 0

𝑃𝑟𝑜𝑝𝑜𝑠𝑒𝑑 𝑚𝑒𝑡ℎ𝑜𝑑 3 𝑄𝑊𝐹∗ < 0

(27)

Fig. 3. Voltage profiles in a radial according to the reactive power control of WTGs

IV. Simulation Results and Discussion

In order to compare the performance of the proposed methods, simulations were

conducted using the wind farm model presented in section 2. The wind speed and the

wind farm reactive power reference are assumed to be changed every 15 min during 24 hours as shown in Fig 4. To take the wake effect into consideration, three different wind

direction scenarios are included in the simulation. The wind direction is defined

according to [23] and displayed in Fig 1. The minimum and maximum voltage limits

of each WTG are set as 0.95 p.u. and 1.05 p.u., respectively. A simple rule is applied

for the OLTC operation. The OLTC tap is adjusted step by step if there is no feasible

solution due to the voltage limits. The simulation process is summarized in the flow

chart of Fig 5. Five allocation methods, i.e. conventional even allocation method and

the proposed methods 1–4, are applied for each simulation scenario. The power loss

and the number of tap changer operation are analyzed.


(a)

(b)

Fig. 4. Simulation scenarios: (a) wind speed, (b) wind farm reactive power reference

Update wind speed and QWF*

Calculate the active power of each WTG

Update busdata and linedata.

NO

Initialization

Power flow calculation

Solve the problem of reactive power

allocation with proposed method

Adjust OLTC

tap positionFeasible Solution?

Update busdata with new reactive power

& Power flow calculation

Yes

n = n + 1

n = 96?

End

Yes

NO

Fig. 5. Simulation flow chart

Fig 6 compares the voltage profiles of the wind farm when the even allocation method

and proposed method 1 are applied. It obviously displays that the voltage profiles along

each redial line under proposed method 1 is flatter than that of even allocation method.

Moreover, the voltage difference among the radials is also decreased significantly. It

means that the OLTC can more easily control the voltages of wind farm within the limit

with less tap change. Table 5 summarizes the tap operation number (NTap) and the power

loss in MWh of five allocation methods considering the effect of wind wake. The tap

operation number of the proposed allocation method 1 was much smaller than that of

0 5 10 15 20 256

8

10

12

14

Time (hour)

Win

d s

peed (

m/s

)

0 5 10 15 20 25-0.4

-0.2

0

0.2

0.4

Time (hour)

QW

F* (

pu)


the even allocation method, as desired. However, at price of it, loss was increased by

53.9% in average. With the proposed method 1, it was observed that some WTGs

supply reactive power while some WTGs in the same radial absorb reactive power to

make the voltage profile flattened. It means that the excessive reactive power

transmission through the cable is the reason for the increased loss of the proposed

method 1.

Fig. 6. Voltage profile under even method and proposed allocation method 1

Table 5. Tap changes and loss under various methods

Wind

direction Parameter

Even

method

Proposed methods

method 1 method 2 method 3 method 4

w/o

wake

NTap 19 14 15 17 15

Loss (MWh) 5.13 7.45 5.67 5.47 5.18

0o NTap 15 10 10 14 12

Loss (MWh) 3.39 5.29 3.79 3.68 3.34

90o NTap 19 10 11 15 13

Loss (MWh) 3.16 4.98 3.66 3.59 3.21

180o NTap 13 10 10 12 10

Loss (MWh) 1.85 3.10 2.16 2.15 1.84

The proposed method 2 showed a great improvement in reducing loss, while the tap

operation number were the same or increased slightly compared to the proposed method

1. It is obvious that, by setting the reactive power limit of WTGs according to the value

of 𝑄𝑊𝐹∗ , unnecessary reactive current, and thus the loss, can be reduced. With the

proposed method 3, more reduction was observed in loss compared to the method 2,

but not as much as expected, whereas the increase of NTap was noticeable in every wind

direction scenario. The additional tap changing operations were observed between

hours 5 and 15, when 𝑄𝑊𝐹∗ , changes in relatively low wind speed condition, as shown

in Fig. 7.

Figure 8 displays the comparison of loss between the proposed methods 2 and 3. It was

observed that the loss of method 2 was smaller than that of method 3 when the value of

𝑄𝑊𝐹∗ is positive, while the situation was reversed for the negative 𝑄𝑊𝐹

∗ condition. The

result provides the idea of method 4, which utilizes the methods 2 or 3 according to the

1 2 3 4 5 6 7 81.03

1.035

1.04

1.045

1.05

Wind turbine number

Voltage (

pu)

1 2 3 4 5 6 7 81.03

1.035

1.04

1.045

1.05

Wint turbine number

Voltage (

pu)

Even-row1

Even-row2

Even-row3

Method 1-row 1

Method 1-row 2

Method 1-row 3


sign of 𝑄𝑊𝐹∗ . In the viewpoint of tap operation number, the performance of method 4

lies between those of method 2 and 3. However, the proposed method showed

significant advantage in terms of loss reduction. The loss of method 4 was reduced by

11.2% and 8.9% compared to method 2 and method 3, respectively. Comparing to the

even allocation method, the tap operation number of method 4 was decreased by

between 20% and 31%, depending on the wind direction, while the loss was almost the

same or even smaller.

Fig. 7. Comparison of tap changing operation between the method 2 and method 3

Fig. 8. Comparison of loss between the method 2 and method 3

To further verify the feasibility and reliability, 100 simulation cases with different wind

farm reactive reference scenarios and wind speed scenarios are conducted. The 10 wind

speed scenarios and 10 reactive power scenarios are shown in Figs 9 and 10,

respectively. The simulation process was the same as shown in Fig 5, but only the wake

effect of wind direction 0 is considered in this simulation. In accordance with the 100

experiment results, the average value of tap operation times and loss of the five

allocation methods are summarized in Table 6. The result shows that the conclusion is

the same as the case presented in Table 5. In other words, the proposed method 4 shows

more advantages on coordinating the two goal of the tap operation reduction and power

loss reduction at the same time.

0 5 10 15 20 258

10

12

14

16

18

20

Time (hour)

Tap p

ositio

n

Method 2

Method 3

0 5 10 15 20 25-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Time (hour)

Po

we

r lo

ss (

pu

)

0 5 10 15 20 25-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

QW

F*

(pu

)

Method 2

Method 3

QWF

*


Fig. 9. Ten wind speed scenarios

Fig. 10. Ten wind farm reactive power reference scenarios

Table 6. Average value of tap operation number and loss of five allocation methods

Items Even Proposed

1 2 3 4

NTap 23.8 18.5 19.2 21.3 19.8

Loss 3.59 5.48 3.99 3.92 3.57

In order to analyze and compare the performance of the proposed methods in more

detail, the tap operation number and loss of the proposed methods were normalized to

those of even allocation method. In other words, the performance based on even

allocation method were regarded as the benchmark value 1.0. The distribution of

normalized tap operation number and loss are shown as box-and-whisker plot in Fig.

11 (a) and (b), respectively. According to the result, it is obvious that method 1 shows

the greatest advantage of reducing the tap operation number by nearly 21% but the

power loss is the worst. Method 2 and method 3 are showing greater advantage of less

power loss at the price of increased tap operation number, compared to the method 1.

Besides, the upper whisker of proposed method 3 in Fig. 11 (a) is even bigger than 1.0,

which means that the tap operation number under part of reactive power and wind speed

scenarios were even larger than that of even allocation method. Although the method 4

is not the best one in terms of tap operation reduction, the value is reduced by nearly

16% compared to even allocation. Moreover, it is worth noting that the loss is the least

compared to other proposed allocation methods and almost equal or even less than that

0 5 10 15 20 256

7

8

9

10

11

12

13

14

Time (hour)

Win

d s

peed (

m/s

)

0 5 10 15 20 25-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

Time (hour)

QW

F*

(pu)


of even allocation method. The smaller box range also shows that the proposed method

4 has stronger effectiveness.

(a)

(b)

Fig. 11. Summary and comparison of 100 simulation cases: (a) distribution of

normalized tap operation number, (b) distribution of normalized loss

V. CONCLUSION

In this paper, reactive power allocation methods of WTGs are developed to improve

the economic effectiveness of the wind farm operation. The main objective is to reduce

the number of OLTC operation by improving the voltage profile within the wind farm

while the loss reduction is also considered. Four methods are developed step by step,

and formulated as a quadratic programming problem. In the formulation, the number of

tap operation and the power loss are not presented in the objective function directly.

Instead, the objectives can be achieved by optimizing the voltage profile of WTGs

terminals. Therefore, it is regarded as an optimization-based coordination approach.

The proposed reactive power allocation method for wind farm can effectively reduce

the operation burden of voltage regulating devices. Moreover, the economic efficiency

of wind farm operation can be improved through the coordinated consideration of

power loss and wind farm voltage profile improvement. The simulation results with

100 scenarios can fully confirm that the proposed reactive power allocation method for

wind farms is effective, feasible, and stable.

Acknowledgements

This research was supported by Korea Electric Power Corporation (Grant number:

R18XA04).

REFERENCES

[1] The World Wind Farms Database. The Wind power published on 27 March

2018. Available online: https://www.thewindpower.net.

[2] Anil G. and Arun S., Challenges of integration of wind power on power system

grid: a review, International Journal of Emerging Technology and Advanced

Engineering, 4(4), 2014, 880-884.

0.6

0.8

1

Method 1 Method 2 Method 3 Method 4

Norm

alized N

Tap

1

1.2

1.4

1.6

Method 1 Method 2 Method 3 Method 4

Norm

alized L

oss


[3] A.K. Pathak, M.P. Sharma and Machesh B., A critical review of voltage and

reactive power management of wind farms, Renewable and Sustainable Energy

Reviews, 51, 2015, 460-471.

[4] Shrikant M., Steffy J. and Ishwari T., Improving low voltage ride-through

capabilities for grid connected wind turbine generator, Proceeding of 4th

International Conference on Advances in Energy Research, Mumbai, India,

2013, 530-540 (54).

[5] C. Feltes, S. Engelhardt, J. Kretschmann, J. Fortmann, F.Koch and I. Erlich,

High voltage ride through of DFIG based wind turbines, Proceeding of Power

and Energy Society General Meeting-Conversion and Delivery of Electrical

Energy in the 21st Century, Pittsburgh, PA, United States, 2008.

[6] J. Zhao, X. Li, J. Hao and J. Lu, Reactive power control of wind farm made up

with doubly fed induction generators in distribution system, Journal of Electric

Power Systems Research, 80(6), 2010, 698-706.

[7] B.S. Chen and Y.Y. Hsu, A minimal harmonic controller for a STATCOM,

IEEE Transactions on Industrial Electronics, 55(2), 2008, 655-664.

[8] Numbi, D.W. Juma and J.L. Munda, Optimal reactive power control in

transmission network with a large wind farm connection, IEEE Africon,

Livingstone, 2011.

[9] E. Diaz-Dorado, C. Carrillo and J. Cidras, Control algorithm for coordinated

reactive power compensation in a wind park, IEEE Trans. Energy Conversion

23(4), 2008, 1064-1072.

[10] Z. Chen, Z. Hao and S. Qin, Centralized reactive power control for a wind

farm under impact of communication delay, International Journal of Control

and Automation 7(2), 2014, 85-98.

[11] Y. Dai, X.D. Liu, Y.X. Ni, F.S. Wen, Z.X. Han, C.M. Shen and Felix F. Wu,

A cost allocation method for reactive power service based on power flow

tracing, Journal of Electric Power Systems Research, 64(1), 2003, 59-65.

[12] T. H. Nguyen, D.-C. Lee, T. L. Van and J.-H. Kang, Coordinated control of

reactive power between STATCOMs and wind farms for PCC voltage

regulation, Journal of Power Electronics, 13(5), 2013, 909-918.

[13] X. Zhang, W. Pan, Y. Liu and D. Xu, Improved grid voltage control strategy

for wind farm with DFIGs connected to distribution networks, Journal of

Power Electronics, 12(3), 2012, 495-502

[14] X. Du, Y. Zhang, Z. Dai, H. Liu, S. Wei and J. Liu, Power distribution

strategy considering active power loss for DFIGs wind farm, Journal of

Power and Energy Engineering, 2(4), 2014, 213-219.

[15] L. G. Meegahapola, S. R. Abbott and D. J. Morrow, Optimal allocation of

distributed reactive power resources under network constraints for system loss

minimization, Proceeding of Power and Energy Society General Meeting, San

Diego, CA, 2011.

[16] B. Zhao, H. Li, M. Wang, Y. Chen, S. Liu, D. Yang, C. Yao, G. Hu and Z.

Chen, An optimal reactive power control strategy for a DFIG-based wind

farm to damp the sub-synchronous oscillation of a power system,

International Journal of Energies, 7(5), 2014, 3086-3103.


[17] J. Kim, H. Lee, B. Lee and Y. C. Kang, Hybrid secondary voltage control

combined with large-Scale wind farms and synchronous generators, Journal of

Electrical Engineering & Technology, 8(2), 2013, 399-405.

[18] O. Holmstrom and N. B. Negra, Survey of reliability of large offshore wind

farms part1: reliability of state-of the-art wind farms, Final Report of Upwind

Project, DTU Wind Energy, Technical University of Denmark, 2007.

[19] T. Ackermann, N. B. Negra, J. Todorovic, and L. Lazaridis, Evaluation of

electrical transmission concepts for large offshore wind farms, Proceeding of

Copenhagen Offshore Wind Conference and Exhibition, Copenhagen,

Denmark, 2005.

[20] M. J. da R. B. Marques, Steady State Analysis of the Interconnection of

Offshore Energy Parks, 2010.

[21] Flo, R. Aardal, Configuration of Large Offshore Wind Farms, 2009.

[22] R. Rakaram, K. S. Kumar and N. Rajasekar, Power system reconfiguration in

a radial distribution network for reducing losses and to improve voltage

profile using modified plant growth simulation algorithm with distributed

generation, Energy Reports 1, 2015, 116-122.

[23] F. G.-Longatt, P. Wall and V. Terzija, Wake effect in wind farm performance:

steady-state and dynamic behavior. renewable energy, 39(1), 2012, 329-338.

https://www.researchgate.net/profile/Nicola_Barberis_Negra

Date post:	04-Jul-2020
Category:	Documents
Upload:	others
View:	10 times
Download:	0 times

Reactive Power Allocation Method in a Wind Farm for ... · Keywords – Wind farm operation,...

Documents